This R03 project aims to dissociate two forms of spatial attention: a declarative form driven by an observer's goal and a procedural form driven by implicit learning. Contrary to existing theories that combine goals and implicit learning into a single source of top-down attention, we hypothesize that they differ qualitatively on several dimensions, including the spatial reference frame, reliance on visual working memory, and task specificity. We propose three lines of behavioral work to support this theoretical framework. First, we investigate the impact of explicit instructions on the magnitude and spatial reference frame of implicitly learned attention. A single-system view predicts that explicit instructions should increase the size of implicitly learned attention, and that explicit instructions should modulate the coordinate system used to code attended locations. A dual- system view predicts no effects of explicit knowledge on implicitly learned attention. Second, we examine the effect of increasing visual working memory load on goal-driven attention and implicitly learned attention. A single-system view predicts that increasing working memory load should impair both forms of attention, but a dual-system view predicts that only goal-driven attention is sensitive to working memory load. Finally, we test the transfer of implicitly learned attention between different tasks. If implicitly learned attention is procedural, then it should not transfer between visual search an change detection or between two types of visual search. But if implicit learning yields a generic change in how spatial locations are weighed, then it should transfer between tasks. Although many studies have shown that attention can be directed to spatial locations based on separate sources (e.g., goals and stimulus saliency), these separate sources of information guide attention in similar ways, often resulting in under-additive effects when multiple sources are available. In contrast, the current proposal suggests that spatial attention is fundamentally divided into multiple forms - declarative and procedural. This distinction, though clear in studies of memory, has not been proposed for attention. Future research should examine the brain basis of the dual-system view in normal adults, the development of the dual systems in children, and the selective impairment of each system in neurological and psychiatric patients.